Google updated its generative AI content guidance on Oct. 1 to require that sites manually fact-check AI output before publishing — including titles, meta descriptions, structured data and image alt text. The change is the strongest documented statement yet from Google on how it expects publishers to handle machine-generated content in Search.

According to Google's documentation changelog, the update added three sentences to the "Focus on accuracy, quality, and relevance" section of its AI content guidance. The new text clarifies that generative models predict likely word sequences rather than retrieve facts, and states it is "critical to manually fact-check and review all AI-generated content for accuracy and trustworthiness before publishing." Google says the revision aligns the page with presentations from its developer events.

The practical effect: AI-generated metadata now sits under the same manual review requirement as article drafts. A review process that only checks body copy can miss fields Google says may surface directly in Search results — meta descriptions, structured data and alt text included.

Gemini links carry UTM tags

A Reddit user spotted UTM parameters on links from Google's Gemini to external websites, and Google Search Advocate John Mueller confirmed he sees them too. Mueller asked whether Gemini passes a referrer sites can already track, then offered to forward details about the referrer's handling to the team — preferably with examples he can reproduce.

Google has not documented the tagging process, so it remains unclear when Gemini appends the parameters, Roger Montti reports. For analytics teams, tagged links offer a way to identify Gemini visits without relying on referrer data. Comparing UTM-tagged sessions against referrer-based Gemini data will show how much traffic each method captures.

Desktop CTR fell, mobile rose in Q2

Tool-signal data — not a Google statement — shows a split in organic click-through rates. Advanced Web Ranking's Q2 2026 CTR report, based on its CTR tool drawing from Search Console data, shows desktop CTR at positions one and two down a combined 5.27 percentage points from Q1, while mobile CTR at those positions rose a combined 6.59 points. That reverses the direction of AWR's Q1 findings.

One measurement caveat the report itself does not mention: the comparison period spans a Search Console logging error that misreported impressions from May 13, 2025 to April 27, 2026. Since CTR divides clicks by impressions, any Q1-to-Q2 comparison in your own Search Console data includes the April 27 fix date. A shift at that boundary may reflect corrected impression counts rather than changed user behavior — worth checking before reading the quarter's CTR movement as a real click pattern change.

AI Mode info monitoring goes global

Google's Robby Stein announced on X on Sept. 28 that AI Mode's information monitoring is rolling out globally to all users. AI Ultra and Pro subscribers already had the feature. Users tell AI Mode what to monitor, Search continuously scans for updates, and the system suggests monitoring tasks during searches. All of Stein's examples focused on local topics and shopping.

The post describes "info monitoring capabilities" — not the information agents introduced for Ultra subscribers in June. Stein's June post said updates would include links to the web, but neither that post nor the Sept. 28 announcement explains how Search selects the sources behind an update. Once the rollout reaches an account, setting up a task relevant to your business shows what an update looks like and whether it carries links — the clearest way for publishers to gauge whether the feature sends traffic.

Judge dismisses Penske and Chegg AI Overviews suits

On Sept. 30, U.S. District Judge Amit Mehta granted Google's motions to dismiss antitrust lawsuits from Penske Media and Chegg over AI Overviews. Mehta found the plaintiffs failed to plausibly allege an agreement in which Google exchanged search traffic for their content, and that they lacked standing to challenge Google's alleged maintenance of its search monopoly. He also held that Penske did not establish that Search and AI Overviews are separate products. The court declined to consider state unjust-enrichment claims.

Mehta acknowledged the court does not treat alleged harms lightly, but wrote that antitrust law is not a substitute for lawmakers addressing the economic upheaval new technology causes. The ruling orders no change to how Google uses publisher content in Search or AI Overviews.

The documentation gap

The throughline across the week: what Google has written down versus what it hasn't. The fact-checking requirement appears in a public changelog and can be built into editorial workflows immediately. Gemini's UTM tags surfaced in a Reddit thread; the monitoring rollout arrived via an executive post on X; the AWR report omits the Search Console error embedded in its comparison window. Documented changes integrate directly into process. Undocumented ones require evidence — a Gemini-tagged session, a live monitoring task — to confirm before acting on them.

Watch next: whether Google documents Gemini's UTM tagging behavior, whether Mehta's dismissal survives appeal, and whether AI Mode monitoring updates carry links to publisher sites at scale.